Getting the DNR

In palliative care, code status can be a preoccupation. Goals of care discussions may be the most common reason for a patient being referred to a palliative care service, but often the purpose of such consultations is overtly or implicitly to “get the DNR.” Pressures from referrers, as well as our own views about the appropriateness of cardiopulmonary resuscitation (CPR), may influence how we present the code status discussion to patients and families. Two papers in this issue of the Journal highlight different approaches to conversations about goals of care.

Many residents, but also more senior clinicians, express strong aversion to performing CPR for our kinds of patients. The great majority of resuscitations, regardless of the patient population, fail to restore a rhythm or lead to recovery of consciousness. CPR is not only regularly a failure but also can seem a grotesquely chaotic procedure, especially when we feel it is not beneficial or is even harmful.

“Getting the DNR” can become a matter of pride for some house officers; residents may be known for the gory descriptions they provide to patients or for other techniques of bullying patients to choose to be DNR. ¹ Patients report being badgered to discuss code status or accept “no code.”^2,3 The frequency of orders to limit life-sustaining technologies in one ICU varied by 15-fold among physicians, demonstrating the huge importance of physician factors in these choices. ⁴

But for the patient who necessarily is unconscious during CPR, the aesthetics of the procedure may be of little concern. Choosing DNR is choosing death in case of a cardiopulmonary arrest.

Other clinician biases are problematic for the goals of care discussion. Physicians are not good judges of what patients consider an acceptable quality of life, the outcome most important to patients in end-of-life decisions. Clinicians tend to rate patient quality of life considerably lower than the patients themselves and to use different criteria for a “good death.” ⁵ (Surrogates and patients likewise differ in their acceptance of functional impairment.^6,7)

In addition to these subtleties about conducting balanced code status discussions, some fundamentals of the goals of care conversation have changed over the past decades. First, while we tend to focus on patients with end-stage conditions who do not benefit from CPR, palliative care teams are increasingly seeing patients with a much higher likelihood of responding favorably to the procedure, including persons in earlier stages of terminal illness, selected patients in intensive care, and especially those with cardiac conditions that have the best outcomes from CPR.

The inpatient CPR discussion, then, requires knowledge about how patients with various conditions of various severities are likely to benefit from CPR: to restore a viable cardiac rhythm, to allow for survival to discharge, and to lead to discharge with reasonable mental capacity and functional status. Similar data is needed for discussing outpatient CPR. Patients clearly change their opinions about CPR when given good information about the procedure: in one study, about 50% of those originally wishing CPR changed their minds to DNR when advised about likely outcomes. ⁸

Considerable variations in knowledge about the success rate of CPR are evident among residents. ⁹ Importantly, faculty and residents at three university-affiliated generalist training programs were no better than chance at predicting which patients would survive resuscitation. ¹⁰

Second, while the data is somewhat confusing, we have some new information about the usual outcomes for in-hospital CPR. In an often-cited study of in-hospital resuscitation, 44% of adults had a spontaneous return of circulation and 17% survived to hospital discharge (34% in cases with ventricular fibrillation). Eighty-six percent of discharged survivors retained a roughly normal cerebral status. ¹¹ Similar results are reported from Canada ¹² and in a U.S. study of the elderly. ¹³ But in another report from a U.S. hospital with a dedicated resuscitation team, the discharge rate was 26%. ¹⁴ In contrast, a Swedish study showed survival to one month or discharge of 40% (61% for those with a shockable rhythm), presumably reflecting different patient selection for the procedure. ¹⁵

In the ICU, a recent Canadian group reported 27% survival to discharge after CPR with 24% one-year survival and 16% survival at five years. ¹⁶ Two other studies showed a survival to discharge of 16%.^17,18 To many patients, such outcomes may seem very well worth the attempt.

Recent outcome data on CPR for patients with cancer or a variety of other conditions commonly seen in palliative care practices are lacking, especially if functional or cognitive outcomes are sought at various stages of the illnesses. In one study at a tertiary referral center, only 8% of patients with “end-stage diseases” survived to discharge. ¹⁴ In a meta-analysis of 35 studies on CPR survival, the major variables predicting poor survival were age over 75, poor functional status, chronic disabling disease, altered mental status, metastatic or hematological malignancy, and “end-stage disease.” ¹⁹ Among lung cancer patients receiving CPR in a medical ICU, the hospital survival rate was 40% with only a 12% 30-day survival rate for patients with stage IV non-small-cell lung cancer. ²⁰ Wallace reported on CPR on cancer patients at MD Anderson, most of whom had hematological malignancy or stem cell transplants; two-thirds died with cardiopulmonary arrest and only 2% survived to discharge. ²¹ But reporting on 22 previous studies on CPR in cancer patients, these authors note mostly no survivors to discharge in studies up to 1991 (with the exception of one reported survival rate of 16%), but discharge rates from 1993 through 2002 of 3.4% to 14.3% (weighted average 6%). Again, these may be reasonable odds for some patients.

Third, the grim cerebral damage associated with many instances of “successful” CPR can be reduced with intentional hypothermia.^22,23 Patients who do not wake up immediately after CPR and who are randomized to receive hypothermia show about a 50% improvement in meaningful survival compared to controls.

Finally, we are refining our approach to presenting the DNR discussion and should be offering CPR to fewer patients. When death is inevitable and imminent, we frankly should not be troubling patients and family members to consent to DNR. In other instances, resuscitation is so unlikely to meet the patient's goals that it should not be offered. In simple terms, do not offer harmful or nonbeneficial procedures. ²⁴ Or as Blinderman suggests, CPR is either (1) an option worth consideration by the patient, (2) something that should be presented as inadvisable, ²⁵ or (3) not offered at all. ²⁶ This approach is best backed up by a hospital policy with guidelines on unilateral DNR orders. And in Blinderman's third case, seek “informed assent,” ²⁷ letting the patient know what you have decided: “CPR is not going to prevent your death and might leave you permanently unconscious on a breathing machine in the intensive care unit, burdening your family with the decision to stop the machine. CPR does not make sense for you so I am entering orders to make sure, if such a catastrophe should arise, that you die peacefully and comfortably.”

This issue of the Journal contains two articles that demonstrate different versions of approaching the DNR discussion. In the Volandes study, two consecutive sets of about 25 surrogate decision makers in the medical ICU were asked four simple questions about their knowledge of CPR, including whether CPR in the ICU leads to hospital survival most of the time, whether being put on a breathing machine usually has complications, and what expected hospital survival is after CPR (where the correct answer is less than 10%, though studies cited above give rates of 16% or higher).

The intervention group viewed a video about the procedure before answering the questions. An earlier version of the video is available at www.acpdecisions.org/ and shows (1) a largely negative presentation on the outcomes of CPR (“frequently does not work,” “cannot eat or talk” while ventilated); and (2) a dramatic and grotesque view of CPR performed on a dummy, including what I register as a clinician shaking his head in dismay.

Not surprisingly, the video produced a modest but significant improvement in surrogate understanding of prognosis (2.0 out of 4 in the nonintervention group versus 2.9 in the intervention cohort). Indeed, given the “state of the science-fiction” in U.S. public knowledge about CPR, ²⁸ both groups did rather well.

Importantly, the authors also studied the impact on code status orders at the time of discharge or death. In previous studies by Volandes, a video of a drooling, barely responsive patient with far-advanced dementia who “cannot talk or communicate understandably” led families to choose comfort care for their demented relative.^29,30 The video in the current study did not significantly change this outcome though the trend was toward more choices for DNR.

While the Volandes study may have provided some rudimentary education and could be considered as part of a fuller discussion about goals of care, it might be viewed as a high-tech version of bullying families to get the DNR, using a one-shot intervention that pays little attention to eliciting patient understanding, values, goals, or preferences.

In the second paper in this issue of the Journal, Szmuilovicz reports on the one-year follow-up of a prospective, randomized controlled trial of multimodality communication skills education for first year internal medicine residents. ³¹ [Disclosure: Dr. Szmulowicz trained as a fellow in the palliative care program I directed.] Educational interventions typically report on immediate or short-term results, often leaving the reader wondering whether sustained improvement occurs; well-designed, longer term follow-up studies are rare. ³² Szmuilowicz now demonstrates consistent improvement in skills over the course of a full year for the intervention group, while the control residents did not improve.

Communication skills training here was not meant to get the DNR but rather to deepen the ability of residents to carry out a skilled goals of care discussion. Residents were taught how to assess the patients' understanding of their condition and prognosis, discuss outcomes of attempted resuscitation, and propose a care plan that respects patient goals, values, and concerns. ³¹

If we value a focus on goals rather than on technical issues in advance care planning, if we identify informed participation and support for patient and family as essential in end-of-life discussions ³³ rather than using shortcuts to get the DNR, if we recognize the important opportunity we have in promoting good communication in association with code status discussion,^25,34,35 the Szmuilowicz intervention is just the type of program that palliative care clinicians should endorse.^36,37 As stated by Ehlenbach,

Communication with patients and their families about resuscitation status should be only one small part of a large discussion about the patient's values and goals of care. … ³⁸

We believe that the only truly effective means of improving patient decision making would be to provide physicians and other clinicians with training that facilitates their ability to uncover a patient's values and goals regarding medical care and that helps patients and their families incorporate these values and goals into decisions regarding CPR. ³⁹